IBM CEO Arvind Krishna details the company's strategic shift to a hybrid cloud and AI software focus, driven by key acquisitions like Red Hat and the divestiture of its IT services business. He offers a nuanced perspective on the AI market, predicting some infrastructure overbuild and commodity large models, while expressing strong confidence in quantum computing's future impact by 2029. Krishna also reflects on fostering a risk-taking culture and the unique challenges and opportunities of leading a legacy tech giant.
Every atomic assertion extracted from the underlying record, ranked by evidence strength.
IBM is largely a hybrid cloud and AI software company.
Some of the AI infrastructure build-out is probably a bit ahead of what the world can tolerate for the next few years.
AI advancements are bigger than mobile and cloud, comparable to the internet.
IBM expects useful quantum computers to be available by 2029.
The biggest thing Arvind Krishna has done as CEO is make IBM's culture much more willing to take risk.
Arvind Krishna has orchestrated one of the most striking turnarounds in big tech since becoming CEO.
Arvind Krishna became the CEO of IBM in 2020.
Arvind Krishna is the chairman and CEO of IBM.
Nikolaj Tangen is the CEO of the Norwegian Southern Wealth Fund.
Value is going to be derived from data, and AI is the only technology known to unlock value from vast amounts of data.
One-third of IBM's revenue is in consulting.
Approximately 20% of IBM's revenue comes from hardware.
Arvind Krishna's diagnosis for IBM's decline was that it was trusted but considered part of the past, not the future.
IBM needed to do things relevant for people's future, even if not immediately the biggest revenue.
IBM's AI focus is on enterprise clients like Nestle, Elevance, Pepsi, and Bank of America, for functions like procurement and accounts payable.
It took another three years after 2019 for the world to wake up to AI's potential.
Arvind Krishna was convinced in 2019 that AI would be big.
Today, IBM is growing faster than it has for a long time.
IBM needs to focus on selling specific capabilities at great prices and quality to smaller clients.
In 2017, Arvind Krishna concluded that IBM becoming a big investor in public cloud was not good for the company.
Arvind Krishna has been with IBM for over 35 years.
IBM would likely still be number five in public cloud after five years of significant investment.
IBM chose to partner with big cloud providers instead of competing directly.
Red Hat was acquired to provide a portfolio that makes IBM a great partner for other cloud providers.
Investing capital into software M&A was considered a better return for IBM.
IBM spun off its IT services business.
The IT services business represented a third of IBM's workforce.
The IT services business was declining at 5%.
IBM wanted to be a company based on innovation, high margins, and growth.
Confluent is considered the best infrastructure for moving and exposing data for AI.
Red Hat's engineering function will likely not be integrated due to its open-source nature.
IBM aims for full integration on the go-to-market side for acquisitions to leverage its global presence.
IBM was a strong believer that sovereignty would remain important for many years to come.
Red Hat is the only acquisition that IBM has not fully integrated yet.
Confluent, with the Kafka backbone, is best in the world at making data real-time.
IBM's culture had become very risk-averse due to years of decline.
A 10 to 15% refreshment rate per year for employees is considered good.
Arvind Krishna believes he has been slow in client expansion, particularly beyond large clients.
IBM doubled down on helping clients transition towards a hybrid cloud.
When Arvind Krishna took over as CEO, IBM had been declining for years.
A gigawatt of power for AI data centers costs $60 billion to $80 billion worth of semiconductors.
People have committed to over 100 gigawatts of AI data center build-out.
Catching up in public cloud would require spending $5 billion to $10 billion a year.
Hybrid cloud and AI software constitute almost half of IBM's total revenue.
Arvind Krishna does not believe the incremental revenue needed for the AI infrastructure build-out is there.
Many of the largest AI models are going to be a commodity.
Commodities have low switching costs, meaning margins will not have a massive moat.
Only two or three companies will likely be able to build the largest AI models and survive.
Companies with very large consumer businesses and natural distribution advantages on the consumer side are likely AI winners.
The enterprise side of AI is wide open for winners.
There is a human timescale for technology adoption, with people taking time to embrace new tech due to risk concerns.
Technology adoption cycles have accelerated over time: computers (20 years), PCs (10 years), internet (5 years).
AI is currently in its "second innings" of adoption.
AI adoption will likely take three to four years to be embraced by enough people.
IBM is one of the most iconic technology companies in the world.
The internet had a massive impact, enabling global business, movement of work, and the rise of social media and cloud.
IBM's previous attempt with Watson in 2011 aimed to prove AI could solve unimaginable problems like natural language questions.
Watson winning Jeopardy proved AI could interpret complex language.
IBM failed with Watson by constructing monolithic applications and picking the hardest vertical, health.
If Watson had focused on helping corporations with customer care or digesting documents, IBM would be five years ahead on AI.